Recent Immigration Has Been Good for Native-Born EmploymentJune 2018

Ryan D. Edwards, Ph.D & Mao-Mei Liu, Ph.D

2 bipartisanpolicy.org

ACKNOWLEDGMENTS

The authors are grateful to the Bipartisan Policy Center for providing financial support for this research via BPC contract 17-IMM-102, and they would like to thank

Theresa Brown, Dennis Feehan, Joshua Goldstein, Hunter Hallman, Carl Mason, Kenneth Megan, Jose Perales, and Cristobal Ramon for providing helpful comments

and advice during the project. All errors and opinions are our own.

AUTHORS Ryan D. EdwardsPh.D. in Economics

Curriculum Coordinator,

Data Science Education Program,

And Research Associate,

Berkeley Population Center,

University of California, Berkeley

ryanedw@berkeley.edu

Mao-Mei LiuPh.D. in Sociology

Lecturer, Sociology Department

University of California, Berkeley,

And Visiting Scholar,

Population Studies and Training Center,

Brown University

mao-mei_ liu@berkeley.edu

3bipartisanpolicy.org

IntroductionGroups in favor of reducing immigration often voice the concern that immigrants compete with native-born Americans for jobs, and that this competition reduces the rates at which native U.S. workers are employed. This perspective often stems from the basic economic law of supply and demand in which an increase in the supply of workers should push down wages or produce unemployment when other things are held constant.

It is true that the laws of supply and demand apply to all participants in the U.S. economy; all residents, businesses, workers, and owners of capital must reckon with the market conditions where they operate. But U.S. workers are not a monolithic group, nor are U.S. businesses, consumers, or immigrants. While many immigrants have low levels of education compared to natives, some possess advanced degrees. Similarly, U.S. workers are a diverse group. Not all U.S. workers compete with immigrant labor; some are coworkers or supervisors who benefit from their presence. In reality, combinations and adjustments in the economy means that the actual impact of immigration on U.S. jobs could be negative, could be zero, or could even be positive if immigration expands new business opportunities by stimulating demand. We need to examine real-world data in order to reveal the true story about the impact of immigration on the U.S. job market.

When we examined trends in employment rates of native U.S. workers compared to trends in foreign-born shares of the local labor force between 2005 and 2016, we found that employment rates for native workers actually rose by a small amount when more immigrants arrived. This pattern held true across a diverse set of U.S. regions and did not reflect any exodus of native workers from the labor force. The data show that the presence of immigrant labor coincided with enhanced employment opportunities for native workers during this period, meaning that the arrival of these individuals does not reduce native employment rates.

These results have significant implications for the policy debate around immigration. While economic vulnerability among U.S. workers and families is a real phenomenon, our finding—especially the fact that immigration does not coincide with lower native employment rates—should lead to a reassessment of policies that seek to restrict immigration because of an assumption that immigrants take jobs from native workers. We believe that policymakers should approach reforms to immigration policy based on the evidence, which does not show significant job effects of immigration.

When we examined trends in employment rates of native U.S. workers compared to

trends in foreign-born shares of the local labor force between 2005 and 2016, we found that

employment rates for native workers actually rose by a small amount when more

immigrants arrived.

4 bipartisanpolicy.org

Basic Economic TheorySometimes policymakers and interest groups who are opposed to immigration voice the concern that immigrant workers will cost Americans jobs, as if there is a fixed and unchanging number of available jobs. In this view, immigrant workers by their very presence must be taking jobs away from native workers. This popular misconception, often called the “Lump of Labor Fallacy,” is implicitly held in many spheres of public policy, but is uniformly rejected by economists across the political spectrum.a A compelling counterexample in the U.S. is the monumental rise of female labor force participation from 32% to 60% between 1950 and 2000, which did not reduce male employment by the same amount over this period.1

Economic theory begins with the reality that workers supply labor in markets, where competing and complementary factors like other similar workers, managers, subordinates, and technology all play important roles. Simple economic theory suggests that without any shift in the demand for workers, an increase in the supply of workers should either reduce wages or raise unemployment. This dynamic is illustrated in Figure 1A below, where w* indicates the original equilibrium wage at which all workers looking for work could find a job, and the labor supply curve has shifted outward to the right because of the arrival of immigrant workers. The new intersection of supply and demand implies that wages should fall, and that if workers cannot afford to work for less than the original wage, unemployment will increase instead.

Figure 1A. An Increase in the Supply of U.S. Workers

This reasoning is sound but potentially incomplete. First, the arrival of immigrant workers could prompt U.S. businesses to invest in new capital, acquiring more equipment, structures, and machinery to complement the additional labor. Second, the arrival of immigrant workers could make native U.S. workers more productive, if immigrants’ skills and talents are more complementary with those of natives rather than being direct copies. For example, the availability of immigrant labor could enhance the productivity of natives through more efficient division of tasks and responsibilities

a For a view from the right, see John F. Cochrane, “Trade and Immigration,” Chapter 9 in George P. Shultz, ed., Blueprint for America (Palo Alto: Hoover Institution, 2016). 109-125. Available at: https://www.hoover.org/sites/default/files/research/docs/george _ shultz _ blueprint _ for _ america _ ch9.pdf. For a view from the left, see Paul Krugman, “Lumps of Labor,” The New York Times, October 7, 2003. Available at: https://www.nytimes.com/2003/10/07/opinion/lumps-of-labor.html.

Unemployment

Demandfor Workers

Supply ofWorkers

w*

Wage

Number of Workers

5bipartisanpolicy.org

between them, playing to the unique strengths of each.b Both of these scenarios suggest that immigration might increase the demand for U.S. workers by making them more productive. Figure 1B shows the two dynamics together, which could either raise or lower native employment and wages, depending on the size of these two effects.

Figure 1B. Increases in the Supply and Demand for U.S. Workers

There are many examples of economic phenomena where simple theory is equivocal. The sea-change in women’s work during the second half of the 20th century is one such example. If only large increases in female labor supply were occurring, the simple model would predict large reductions in wages for men and for subsequent younger cohorts of women. History has shown the reverse to be the case. The demand for labor must have shifted strongly outward due to advancing technology and ostensibly to new complementarities in production between male and female workers.

Another famous example is the case of the college wage premium, which is a persistently large bonus in the earnings of U.S. workers who hold a four-year college degree compared to other workers.c Despite sustained increases both in the absolute number and relative share of workers with a college degree over the past several decades, the college premium remains large. This pattern implies that a significant increase in the demand for college-educated workers must have coincided with the expansion in their supply, and it must be continuing.2

Simple economic theory alone cannot reveal or model these nuances of reality. Rather, it is the application and testing of a variety of theoretical perspectives with observed patterns in the data that reveal the truth.

Before we describe our analytical framework and our results, we discuss the data that we examine and the broad contours of recent patterns in U.S. immigration that they reveal.

b For a discussion of capital investment behavior by firms and worker complementarity, see Ryan Edwards and Francesc Ortega, “The Economic Impacts of Removing Unauthorized Immigrant Workers,” Center for American Progress Report 2016. Available at: https://www.americanprogress.org/issues/immigration/reports/2016/09/21/144363/the-economic-impacts-of-removing-unauthorized-immigrant-workers/. And Ryan Edwards and Francesc Ortega, “The economic contribution of unauthorized workers: An industry analysis,” Regional Science and Urban Economics 67(2017): 119–134. Available at: http://dx.doi.org/10.1016/j.regsciurbeco.2017.09.004.

c For background and a discussion of recent trends, see Robert G. Valletta, “Recent Flattening in the Higher Education Wage Premium: Polarization, Skill Downgrading, or Both?” in Charles R. Hulten and Valerie A. Ramey, eds., Education, Skills, and Technical Change: Implications for Future US GDP Growth (Chicago: University of Chicago Press, 2016). Available at: http://www.nber.org/chapters/c13705.pdf.

Number of Workers

Demandfor Workers

Supply ofWorkers

w*

Wage

6 bipartisanpolicy.org

Recent Trends in ImmigrationThe U.S. is a nation of immigrants, both in terms of its historical origins and trends in the share of the population who are immigrants. In 2016, nearly 44 million U.S. residents were foreign-born, representing 13.5 percent of the population. This is near the historical peaks of the late 19th and early 20th centuries, as shown in Figure 2A.

Figure 2B spotlights trends in recent data, especially the annual iterations of the American Community Survey (ACS) from 2005 through 2016, the most recent public release. The ACS was introduced by the U.S. Census Bureau as a replacement for the long form of the decennial census, and it contains detailed characteristics that are measured annually for about 1 percent of the U.S. population. As Figure 2B depicts, the immigrant share of the population actually plateaued and fell slightly during the Great Recession of 2007-2009 before resuming a more gradual upward climb.

Note: Statistics are drawn from the decennial census up to 2000 and the American Community Survey from 2005 onward.d

Underneath these national trends are strikingly different levels and trends in the immigrant share across U.S. geographic regions. We can most easily visualize these when we focus on the nine geographic divisions defined by the U.S. Census Bureau.e Figure 3 displays the foreign-born shares of the labor force, consisting of all employed workers plus the unemployed, for each U.S. division in 2010. Although immigrants as a group differ from natives in terms of their age structure and are more likely to be of working age, regional patterns in the percent of foreign-born are similar whether we examine the total population or the labor force. Because we are concerned with labor market impacts in this study, we focus on the labor force.

d Historical statistics are provided by Mary C. Waters and Marisa Gerstein Pineau, eds., The Integration of Immigrants into American Society, Panel on the Integration of Immigrants into American Society, Committee on Population, Division of Behavioral and Social Sciences and Education, National Academies of Sciences, Engineering, and Medicine (Washington: National Academies Press, 2015). Available at: https://www.nap.edu/catalog/21746/the-integration-of-immigrants-into-american-society; Annual data from 2005-2016 are derived by the authors from the American Community Survey extracts provided by Steven Ruggles, Katie Genadek, Ronald Goeken, Josiah Grover, and Matthew Sobek, Integrated Public Use Microdata Series: Version 7.0 [dataset]. (Minneapolis: University of Minnesota, 2018). Available at: https://doi.org/10.18128/D010.V7.0.

e The Census Bureau graphically depicts regions and divisions in a map of the 50 states here: https://www2.census.gov/geo/pdfs/maps-data/maps/reference/us _ regdiv.pdf

0%

2%

4%

6%

1850

1870

1860

1880

1900

1890

1910

1920

1940

1930

1950

1960

1980

1970

1990

2000

2010

2020

12%

8%

10%

Year

14%

16%

Decennial Census ACS (2005-2016)

0%

5%

10%

Year

15%

Decennial Census ACS (2005-2016)

1980 19981989 2007 2016

Figure 2A. The foreign-born share of the U.S. population since 1850

Figure 2B. The foreign-born share of the U.S. population since 1980

7bipartisanpolicy.org

Figure 3. Foreign-born shares of the labor force across U.S. regions, 2010

Note: Statistics are drawn from the American Community Survey and displayed over original map produced by the U.S. Census Bureau.

The lowest immigrant share of the labor force across U.S. divisions in 2010 was 5 percent in the East South Central division, which consists of Alabama, Kentucky, Mississippi, and Tennessee. The highest share was 29 percent, or almost six times larger, in the Pacific division, which includes Alaska, California, Hawaii, Oregon, and Washington State. Between these extremes, the immigrant share in the other seven division still varied broadly, ranging from 6 percent in the West North Central division, which includes Iowa, Kansas, Minnesota, Missouri, Nebraska, and North and South Dakota; to 21 percent in the Middle Atlantic division, consisting of New Jersey, New York, and Pennsylvania.

Figure 4. Foreign-born shares of the labor force across U.S. divisions, 2005 and 2016

Note: Statistics are drawn from the American Community Survey.

Pacific: 30% Mountain: 14%

WEST

West South Central: 16%

East South Central: 5%

South Atlantic: 16%

West North Central: 6%

East North Central: 9%

Middle Atlantic: 21%

1–10%

Immigrant share of the labor force

11–20%

21–30%

New England: 14%

NORTHEAST

0%

15%

10%

5%

25%

20%

35%

30%

NewEngland

MiddleAtlantic

EastNorth

Central

WestNorth

Central

SouthAtlantic

EastSouth

Central

WestSouth

Central

MountainPacific

2005 2016

8 bipartisanpolicy.org

In addition to these large differences in the immigrant share across geographic divisions, there also are large differences across divisions in how those shares are changing over time. As shown in Figure 4, the Pacific division maintained the largest share of immigrants in its labor force over the period, which was virtually unchanged at 29 percent. But other divisions with lower immigrant shares experienced large increases. Absolute percentage-point increases, visible in the differences between the two bars for each division, were largest for the Middle Atlantic division, and for the West South Central division. Those divisions began the period at 19 percent and 15 percent, respectively, and ended at 22 percent and 17 percent.

For divisions less historically accustomed to immigration like the West North Central and East South Central divisions, increases were smaller in absolute terms but very large on a proportional basis. The percent of foreign-born in the West North Central division rose 40 percent from 5 percent to 7 percent of the labor force. The East South Central division began from a smaller base share of 4 percent and experienced a similarly large percentage increase. In summary, Figure 4 reveals there are great asymmetries across the country in recent levels and changes in immigration.

These recent patterns of immigration have sometimes been called the “new geography of immigration.”3 This dynamic is often thought to be associated with an increase in economic production in suburban as opposed to dense urban areas. Whatever the cause or correlates might be, this new geography of U.S. immigration implies that areas of the country that currently receive the most immigrants do not benefit from the historical experiences of traditional immigrant-receiving areas. And if the new geography of immigration is indeed driven by the growth of employment opportunities in suburban areas, one may expect that native workers, employers, and industries in the new receiving areas are keenly aware of the presence of new immigrants and may be interested in assessing their role in the changing economy.

As we have discussed, simple economic theory is unclear on this question, and providing a full answer requires examining the data. In the next section, we explain our analytical framework and estimation approach and discuss how it mirrors the state-of-the-art in the relevant scholarly research.

9bipartisanpolicy.org

Analytical ApproachIf it were possible, the ideal way to assess the effect of immigration on native employment would be to examine a labor market before and after an influx of immigrant workers, and then compare what happened to a counterfactual case where the very same labor market was observed during the same period but without the new immigrant workers. Because this ideal comparison cannot be observed, social scientists usually seek instead to compare changes they can observe over time among labor markets that fit the characteristics of control and treatment groups, just like in a randomized medical trial.

The most famous example of a study comparing control and treatment groups in the field of immigration research is the comparison of the Miami labor market around 1980, which received a large number of Cuban “Marielitos” as refugees from the Castro regime, to a set of reasonable “control” cities during the same period.4 By failing to find evidence of any reduction in local wages or employment in Miami relative to other cities, the original study turned conventional wisdom on its head. Academic controversy about this result continues, with much focus on narrowly defined groups of vulnerable native workers and whether the underlying data are rich enough to accurately measure their conditions.5 Although disagreements remain, the consensus in academic thinking about the economic impacts of immigration on natives is that effects are usually small or zero and tend to vary across native characteristics.f

In a new study, we revisit this question by comparing a broad array of geographic labor markets “treated” with increased immigration to another broad array of labor markets that are not.6 We use a standard statistical estimation technique and we apply it to a relatively new and rich dataset, the 12 publicly available annual waves of the American Community Survey (ACS).g Our technique, which is commonly used in applied social science literature, is a generalization of the method that compares changes over time among treatment and control groups to estimate the effect of the treatment, which in this case is higher immigration.

Each annual wave of the ACS provides roughly 3 million observations of residents in households. Our unit of analysis is the Public Use Microdata Area (PUMA), of which there are roughly 1,000 that do not cross state or regional boundaries. PUMAs are similar to counties (which number approximately 3,000), but are larger than the smallest counties by population. The Census Bureau designed the PUMAs in order to capture the highest geographic resolution annually, while preserving the anonymity of survey respondents.

For our study, we examine PUMAs rather than counties or other levels of geography in order to examine dynamics at the smallest geographic level possible while preserving critical variation over time and maintaining a national scope to our study. Within each PUMA, we examine how the employment rate among native workers changed with the immigrant share of the labor force. Our standard estimation technique then compares changes in the native employment rate within geographic labor markets that are treated by more immigration to changes in native employment within markets where there are different or no changes in immigration. By holding constant other measured influences on employment, the technique then ascribes the observed difference to immigration.

f Two reports of expert panels convened by the National Academies of Sciences on the subject speak with largely the same voice on the state of the scientific literature: James P. Smith and Barry Edmonston, eds., The New Americans: Economic, Demographic, and Fiscal Effects of Immigration, Panel on the Demographic and Economic Impacts of Immigration, Committee on Population and Committee on National Statistics Commission on Behavioral and Social Sciences and Education, National Research Council, (Washington: National Academy Press, 1997). Available at: https://www.nap.edu/catalog/5779/the-new-americans-economic-demographic-and-fiscal-effects-of-immigration. And Francine D. Blau and Christopher Mackie, eds., The Economic and Fiscal Consequences of Immigration, National Academies of Sciences, Engineering, and Medicine. (Washington: National Academies Press, 2017). Available at: https://www.nap.edu/catalog/23550/the-economic-and-fiscal-consequences-of-immigration.

g The method is panel fixed effects, which is a generalization of the difference-in-differences estimation used by Card, ibid. and many others. Details are discussed in Ryan D. Edwards and Mao-Mei Liu, “A new look at immigration and employment in the U.S. since 2005,”UC Berkeley: Center on the Economics and Demography of Aging, 2018. Available at: https://escholarship.org/uc/item/0q21g0b8.

bipartisanpolicy.org

METHODOLOGICAL DETAILS IN BRIEFWe report results based on a standard model in social science that compares changes over time in native employment rates (Y) observed in small geographic areas (PUMAs) with high immigrant labor force shares (X) versus those observed in areas with low immigrant labor force shares. The technical name for this model is a panel regression with PUMA fixed effects, and it is a generalization of the difference-in-differences approach described in the text.

In this approach, we follow the standard procedure of controlling for other PUMA-level variables (Z) that might also affect native employment rates and might also be changing. These include the age structure of the population; percent male; percent in five race/ethnicity categories, including percent Latino or Hispanic; percent in seven educational attainment categories; percent living in a metro area; and indicator variables for each year in the sample (2005-2016).

This method identifies an effect of the immigrant labor force shares (X) associated with the native employment rate (Y) that is analogous to what a researcher would find when comparing two otherwise identical labor markets, one of which received an influx of immigrants while the other did not.

Native Employment Climbs with ImmigrationContrary to what the basic economic theory of immigration predicts, we find that rising foreign-born shares of the local labor force are associated with increases in native employment rates over the 2005-2016 time period. Our model predicts that with every percentage point increase in the foreign-share of the labor force, native employment rate will rise by between 0.055 to 0.075 percentage points, a small but statistically significant effect on local labor markets that is economically meaningful.

In the average sample area, 1 percent of a PUMA’s labor force is about 1,500 workers, while 1 percent of native employment is roughly 1,100 jobs (the average PUMA has a ratio of immigrant to native workers of four to 21). Using the midpoint of our estimated effects, a coefficient of 0.065, we find that an increase of 20 immigrant workers is associated with about one native job created. This result is noteworthy because

it stands in stark contrast to the dire predictions of a simple model in which immigrant workers displace natives.

It is also remarkable because it implies that labor force responses to immigration consist of more than just a keep-working or stop-working response by natives who are already working. Our central result suggests that immigrant and native workers combine in the job market in productive ways that can expand the jobs available. For example, the rapid expansion of hires in a workplace might cause the creation of new management or human resources positions that would not have existed but for the presence of the additional workers. Immigrant workers are diverse, with many varying levels of education and skills, but patterns in the employment of immigrants by occupation reveal that language skills are variable and important in how businesses allocate different types of workers to different tasks.7 Firms might approach the challenge of heterogeneous skills among workers as an opportunity, by combining the unique services of immigrant workers with the unique skills of native-born workers.

Our central result suggests that immigrant

and native workers combine in the job

market in productive ways that can

expand the jobs available.

10

bipartisanpolicy.org

UNEMPLOYMENT AND EMPLOYMENT RATES AND EMPLOYMENT-TO-POPULATION RATIOSThe official unemployment rate, reported at 4.1 percent in March 2018, is derived as the ratio of the unemployed to the sum of employed workers plus the unemployed. Individuals are defined to be unemployed if they are not currently working but are actively looking for work. The denominator in this ratio, the sum of all employed workers plus the unemployed, is called the labor force.

In this study, we examine the employment rate, which equals 100 percent minus the unemployment rate and measures the share of the labor force that is employed. Again, the denominator is the labor force.

A similar measure is the employment-to-population ratio, which is constructed as the ratio of employed workers to the total population. The denominator in this statistic is much broader and is comprised of a) the labor force; b) discouraged workers who are no longer looking for work; and c) students, retirees, homemakers, the disabled, and others not engaged in formal labor markets.

Because discouraged workers and other people are excluded from one measure and not the other, the two are often different. During the Great Recession of 2007-2009, both the employment rate and the employment-to-population ratio declined precipitously at first. In the years since, the employment rate has fully returned to pre-recession levels of about 96 percent, but the employment-to-population ratio has not and currently hovers at 60.4 percent or roughly the level it was in 1986.

ROBUSTNESS OF THE RESULTWe ran our results through a wide range of specification checks to ensure the main findings endured. Overall, our results are robust. We found that our results were stronger in geographic areas with higher population density than in less dense geographic areas, but there was a small positive response of native employment to immigration in both types of areas. This implies that our estimate of the national effect of immigration based on trends across geographic areas was a slight underestimate of the true average effect on native U.S. residents as a whole. This is because treating geographic areas as the unit of observation without accounting for their population differences weights the experiences of denser urban areas the same as the experiences of less dense rural areas. The unweighted results are certainly still relevant, because our politics are sometimes weighted toward the interests of states and other geographies, irrespective of population density. But because of stronger effects in higher population areas, more Americans overall are seeing these positive employment effects.

A major concern we had was whether our primary outcome measure, the employment rate of native workers, sufficiently captured all potential effects of immigration on native workers. As discussed in the accompanying box, the employment rate is the share of the labor force with a job. A more familiar measure is the unemployment rate, which is the share of adults who do not hold a job and are actively seeking work. A potential problem with using this type of outcome measure is the possibility of native workers becoming discouraged and dropping out of the labor force because of competition from immigrants. These workers would not appear in either the employment or the unemployment rate, since they are no longer considered part of the labor force. To explore this possibility, we re-estimated models with the employment-to-population ratio, a measure that would include the discouraged workers. If immigration discouraged native workers from seeking work, the native employment-to-population ratio would fall and thus reveal new information. But our results using the employment-to-population ratio revealed no evidence that immigration was pushing native workers out of the labor force. More immigrants were associated with more native employment, measured either as a share of the labor force or of the whole population.

We were also concerned whether the inclusion of the Great Recession in the time period spanned by our data somehow artificially created the results we found or meant that they implied something different than what they appeared to imply. The time period of our analysis, 2005 to 2016, was tumultuous and eventful, and it included the boom years before the Great Recession, the Great Recession itself, and the prolonged

11

12 bipartisanpolicy.org

recovery. We were most concerned that the recession may be driving the result “in reverse,” namely showing us reductions in immigrant labor and in native employment, and both likely in response to the recession itself and not because of one another. We found this was not the case, and that the main driver of the results was coming from periods of growth in both immigration and in native employment. In particular, it appeared that the era of expansion prior to the Great Recession was a time during which native employment appeared to expand strongly alongside immigration.

Past theory and evidence propose that the effect of immigration will vary for different groups of native workers, and to explore this we re-estimated our model using labor market outcomes for different groups of native workers defined by their educational attainment. We found that the positive effect of immigration was stronger among natives with more education, while the effect was statistically insignificant among natives with less education. The largest positive effect of immigration was on the employment of native workers who had some college education but not a four-year college degree, which we found to be an interesting result.

We also explored how our result varied across the nine census divisions. What emerged is a snapshot of considerable diversity, which is depicted in Figure 5. We found large and statistically significant influences of immigration on native employment in five of the nine division, which were a mix of areas with low and high immigrant shares alike. As Figure 5 shows, we found positive coefficients greater than 0.090 and statistically significant for:

• The Mountain division (Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, and Wyoming);

• The West South Central division (Arkansas, Louisiana, Oklahoma, and Texas); and

• The East North Central division (Illinois, Indiana, Michigan, Ohio, and Wisconsin).

In these three divisions, the coefficient hovered around 0.094.

We also found large, positive effects of immigration for:

• The West North Central division (Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, and South Dakota); and

• The Pacific division (Arizona, Pacific region (Alaska, California, Hawaii, Oregon, and Washington).

The coefficient was 0.079 in the West North Central division, where immigration was very low but grew rapidly during the period; and it was 0.055 in the Pacific division, where immigration was the highest and remained high.

13bipartisanpolicy.org

In the four other divisions, the estimated effect of immigration was statistically insignificant, positively signed in three cases and negatively signed in the New England division. A noteworthy result is the null effect of immigration in the Middle Atlantic division, comprising New Jersey, New York, and Pennsylvania and containing the second-highest share of foreign-born workers of all divisions.

As we have seen, there is great variation in both the levels and changes in the immigrant share across divisions. However, we found that the size of the positive employment effect was not explained by a division’s immigrant share, nor was it explained by how fast the immigrant share was changing. We suspect that differences in the native employment effect across divisions may be associated with the geographic variation in industries and the differential changes in the health of those industries over time. But a complete view awaits future investigation.

Figure 5. Diversity of Model Results Across U.S. Divisions

Note: Each bar represents a coefficient from a different regression of the native-born employment rate on the immigrant share of the labor force. Statistically significant results are represented in blue, while results that lack statistical significance are in white.

The differences across geographic areas in the effect of immigration on native employment is an opportunity for us to expand our understanding of the dynamics of immigrant absorption. It is not a simple story about positive effects on natives in traditional immigrant-receiving areas and negative or zero effects elsewhere, nor is it the reverse. However, given our findings that immigration does have (mostly positive) labor market impacts on native-born Americans, further research on these geographic differences would be welcome.

-0.04

Pacific

Mountain

West South Central

East South Central

South Atlantic

West North Central

East North Central

Middle Atlantic

New England

U.S. Average

0-0.02 0.02

Regression Coefficient

0.04 0.08 0.120.10.06

14 bipartisanpolicy.org

Conclusion: U.S. Immigration and Native Employment Have Risen TogetherThis study reviewed U.S. Census Bureau data to examine how changes over time in immigrant shares of the labor force is or is not related to changes over time in the employment rates of native workers within 1,000 local labor markets defined by geography. Simple economic theory suggests that without any offsetting influences, increases in the foreign-born share of the local labor force that are driven by increases in the supply of immigrant workers might reduce employment rates among native workers.

In stark contrast to this prediction, we find robust evidence that between 2005 and 2016, employment rates among native workers rose when the immigrant share of the local labor force increased. Immigrant workers did not displace natives from jobs, as might be predicted by a simple model in which all workers were identical and businesses did not shift strategies to employ workers productively. Rather, our analysis suggests either: 1) that native workers combined with new immigrant labor in productive ways that created more employment opportunities for natives, such as leveraging new divisions of labor; 2) that businesses expanded by outfitting their workers with more equipment and machinery; or 3) that both may have occurred.

Perhaps most surprising and compelling was the broad-based geographic robustness of this result. Areas with already high shares of immigrant labor were not the only areas that benefitted from the arrival of new immigrants. Instead, we found evidence of substantial employment gains by native U.S. workers in traditionally immigrant-scarce regions as well as in immigrant-plentiful regions.

The size of the effect we estimated was substantial but not enormous. Our estimates imply that every increase of 20 immigrant workers was associated with one additional job held by a native-born worker during the sample period. Native U.S. workers with some college experience but less than a four-year college degree appeared to benefit the most in terms of employment from the presence of new immigrants. The one-to-20 ratio of new jobs for natives to new immigrant workers seems consistent with a story of complementarities between classes of workers that other studies appear to confirm. The data suggest that in recent times, contrary to the popular political narrative, the presence of immigrant workers has not been a threat to the jobs of native-born U.S. workers, but instead a source of modest but real employment gains.

Immigrant workers did not displace natives from jobs, as might be predicted by a simple

model in which all workers were identical and businesses did not shift strategies to

employ workers productively.

15bipartisanpolicy.org

Endnotes1 John F. Cochrane, “Trade and Immigration,” Chapter 9 in George P. Shultz, ed., Blueprint for America (Palo Alto: Hoover Institution, 2016). 109-125. Available at:

https://www.hoover.org/sites/default/files/research/docs/george_shultz_blueprint_for_america_ch9.pdf

2 Charles I. Jones, Macroeconomics, 4th ed. Chapter 7, (New York: Norton, 2016).

3 Audrey Singer, “The Rise of New Immigrant Gateways,” Brookings Institution, Living Cities Census Series, February 2004. Available at: https://www.brookings.edu/wp-content/uploads/2016/06/20040301_gateways.pdf. Audrey Singer. “The New Geography of United States Immigration,” Brookings Immigration Series, No. 3, July 2009. Available at: https://www.brookings.edu/wp-content/uploads/2016/06/07_immigration_geography_singer.pdf. Douglas S. Massey, ed., New faces in new places: The changing geography of American immigration, Russell Sage Foundation, 2018.

4 David Card, “The Impact of the Mariel Boatlift on the Miami Labor Market,” Industrial and Labor Relations Review 43(2), (1990): 245-257. George J. Borjas, “The Wage Impact of the Marielitos: A Reappraisal,” ILR Review, vol. 70(5), (2017): 1077-1110. Available at: http://doi.org/10.1177/0019793917692945. Giovanni Peri and Vasil Yasenov, “The Labor Market Effects of a Refugee Wave: Applying the Synthetic Control Method to the Mariel Boatlift,” NBER Working Paper 21801, (2015). Available at: http://www.nber.org/papers/w21801. Michael A. Clemens and Jennifer Hunt, “The Labor Market Effects of Refugee Waves: Reconciling Conflicting Results,” NBER Working Paper 23433, (2017). Available at: http://www.nber.org/papers/w23433.

5 See especially Borjas, and Clemens and Hunt, supra.

6 Ryan D. Edwards and Mao-Mei Liu, “A new look at immigration and employment in the U.S. since 2005,” UC Berkeley: Center on the Economics and Demography of Aging, 2018. Available at: https://escholarship.org/uc/item/0q21g0b8.

7 Kenneth Megan, “Immigration and the Labor Force, Part II,” Bipartisan Policy Center Blog Post, September 21, 2015. Available at: https://bipartisanpolicy.org/blog/immigration-and-the-labor-force-part-ii/.

16 bipartisanpolicy.org

Notes

17bipartisanpolicy.org

Notes

18 bipartisanpolicy.org

Notes

19

BPC POLICY AREAS

Economy

Education

Energy

Evidence

Finance

Governance

Health

Housing

Immigration

Infrastructure

National Security

@BPC_Bipartisan

facebook.com/BipartisanPolicyCenter

instagram.com/BPC_Bipartisan

The Bipartisan Policy Center is a non-profit organization that combines the best ideas from both parties to promote health, security, and opportunity for all Americans. BPC drives principled and politically viable policy solutions through the power of rigorous analysis, painstaking negotiation, and aggressive advocacy.

bipartisanpolicy.org | 202-204-24001225 Eye Street NW, Suite 1000 Washington, D.C. 20005

1225 Eye Street NW, Suite 1000 | Washington, D.C. 20005202-204-2400| bipartisanpolicy.org